Friday December 16, 2005
Friday December 16, 2005
What is SWaP and iGenCPU SWaP values
- Performance: Using industry-standard
benchmarks or your own benchmark !
- Space: Measuring the height of the server in rack units (RUs).
- Power: Determining the watts consumed by the system, using data from actual benchmark runs or vendor site planning guides
The SWaP metric is calculated this way :
I recently provided iGenCPU v2.1 benchmark results for various platforms (see previous entries). I did start with this benchmark as it is the absolute worst case for the T2000. Let see how it translates into SWaP numbers :
Or as expressed in a chart :
So what is the message here ?
If you are running floating point intensive applications (the immense majority of commercial applications are not), and you need a small form factor, the AMD equipped V20Z/V40z or better the Galaxy line ( Sun Fire™ X4100 and Sun Fire™ X4200 ) are the right answer.
I hope I convinced you here that the SWaP metric was not designed to make the UltraSPARC-T1 a winner every time but to provide a critical metric for modern datacenters !
In 2006, we will look at a second microbenchmark called iGenRAM , and then explore LDAP performance (iGenLDAP), database performance (iGenOLTP) and even web performance (iGenWEB).
I am leaving for Tahoe, so see you on the Heavenly slopes or next year on this forum...
Dec 16 2005, 11:54:42 AM PST Permalink
Comments:
SWaP as described above is an absurd metric (even though I'm completely FOR using space and power consumption as parts of the metric);
useless for anything but marketing wars. what's the supposed meaning of Space*PowerConsumption ?
Simple - but much more reasonable imho - formula would be
Performance/(Price + c1*Space + c2*PowerConsumption)
where constant c1 reflects cost of space and c2 - cost of electricity (including cooling) in a particular datacenter.
Posted by Igor on December 19, 2005 at 11:45 AM PST #
Post a Comment:
Comments are closed for this entry.